On 2009-04-06 00:53-0700 Alan W. Irwin wrote: > Hi Alban: > > I have a contact who was kind enough to put together a simple Qt application > for me that wrote unicode text to a GUI text box. I changed that application > (by brute force since my C++ skills are not the best) to specifically use > the text string > > r(theta)=sin 5theta, > > where "theta" in the above equation stands for the actual Greek letter, > unicode 0x398. The point of this example was to prepare a simple test case > for a Qt bug report. I was fully expecting to see a misalignment of the > theta relative to the Latin letters just like we get in the title to Plplot > example 3 with any of the qt devices. However, that expectation was wrong! > In fact, the simple Qt text application lines up the theta perfectly both > with Qt-4.4.3 provided by Debian Lenny, and the pre-release version of > Qt-4.5.0 that you can download from trolltech. > > I have attached this simple test case as a tarball. To build it, unpack the > tarball, cd to the created directory, configure the application > by running the "qmake" command, and then run "make" to > generate a Qt executable called greektest. When you run that application > you will see a nicely aligned result unlike the title rendered by all qt > devices for the 3rd PLplot example. (I have also attached the result > I get here for -dev pngqt for example 3 which demonstrates the alignment > problem.) > > Because I cannot reproduce the alignment issue with a simple test case, I am > beginning to feel there are still text alignment bugs lurking in our qt > device driver that affect only a small subset of glyphs in our examples > (notably the Greek glyphs and a few more). > > I plan to change the text attributes in the simple example to be as > consistent as possible with the qt device driver. The approach I am taking > here is to find a way to replicate the misalignment issue with a simple test > case. Once I know how to change the text case to misalign the characters, > it should be a small step to do the opposite and fix the qt device driver to > align the characters. > > I will try to deal with this issue in the suggested way before the release, > but I have other release-related issues on my plate so my time is limited, > and I may not get to it before the release. Also, my approach might not work > to find the source of the problem. Thus, I encourage you to work on this > issue as well (or completely take responsibility for fixing the issue if > it looks like it would be easy for you). > > If you are agreeable to working on this issue, the first question is can you > reproduce the misaligned Greek characters in the title of PLplot example 3 > (the theta is roughly half a character height too high on my platform as you > can see from the second attachment). Assuming you verify the issue, you may > want to take a different debugging approach than the one I discussed above > such as experimenting directly with text attributes for the qt device driver > until you find a combination that gives good alignment for our example 3 > title. If you can do that, I would be happy to follow up with comprehensive > tests for all our examples to make sure all known character alignment issues > are fixed by your change and there are no alignment regressions.
To everyone here who would like to help out with testing: To help understand what I suspect is a qt device driver character alignment issue, I would appreciate some specific testing of example 3 using -dev pngqt to see if there is any platform/set of fonts where you get good character alignment unlike the pngqt result for example 3 that I attached earlier in this thread. Alan ------------------------------------------------------------------------------ This SF.net email is sponsored by: High Quality Requirements in a Collaborative Environment. Download a free trial of Rational Requirements Composer Now! http://p.sf.net/sfu/www-ibm-com _______________________________________________ Plplot-devel mailing list Plplot-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/plplot-devel
